An Image Mosaic Method Based on SIFT Feature Matching

Abstract

This paper concerns the problem of image mosaic. An image matching method based on SIFT features and an image blending method of improved Hat function are proposed in the paper. SIFT feature is local feature and keeps invariant to scale zoom, rotation and illumination. It is also insensitive to noise, view point changing and so on. Because of this our method is insensitive to orientation, scale and illumination of input images, so it’s possible to accomplish image mosaic between arbitrary matching images and the Hat function blending algorithm with global intensity revise makes the mosaic image accepted by human eyes.